A widely circulated commentary argues that merchants systematically undercount the true cost of accepting crypto. A close reading of its own sourcing suggests the argument is directionally right and evidentially hollow.
A commentary piece examining the operational costs of crypto payment acceptance has circulated through payments and e-commerce circles in recent weeks, making a case that has genuine merit: that merchants comparing a crypto gateway’s headline processing fee against a card network’s merchant discount rate are comparing the wrong numbers. The real cost, the argument goes, sits in failed transactions, misdirected deposits, manual refund workflows and customer support hours — none of which appear on a rate card.
The problem is not the thesis. It is the evidence assembled to support it, and the identity of the party that supplied most of that evidence.
The Thesis: From Fee Comparison to Total Cost of Ownership
The core argument is that crypto payments should be evaluated on total cost of ownership rather than headline rate. On-chain settlement is cheap and increasingly reliable; the friction lives at the boundary where an irreversible ledger meets a commercial system that requires reversibility. Refunds, chargebacks by another name, misapplied destination tags and wrong-chain deposits all require human intervention, and human intervention has a price per hour that no published fee schedule captures.
That framing is sound, and it is overdue. Payment acceptance has always carried costs that sit outside the discount rate — chargeback fees of roughly $15 to $25 per dispute in card payments are a well-known example — and crypto acceptance introduces a different, less familiar set of them.
But a framework is only as useful as the data poured into it, and here the commentary runs into trouble.
Where the Numbers Come From
Of the roughly two dozen discrete factual claims in the piece, the most load-bearing ones fall into two categories: figures attributed to unnamed research, and figures supplied directly by the payment gateway the article ultimately recommends.
The unnamed category includes a claim that 37 percent of institutions say each failed payment costs more than $20, attributed to a study of 150 executives with no named research house, no publication link and no disclosed sample composition. It includes a 3.7-day median refund interval drawn from an unnamed dataset with no stated sample size, currency mix or transaction size distribution. And it includes a comparison of 21 minutes versus 4.3 hours for some aspect of refund handling, drawn from a two-and-a-half-month window with no disclosed sample.
The named category is more revealing. A reported 103.8 percent year-over-year increase in integrations and a 92 percent increase in partners completing a first transaction both trace to the gateway’s own social media posts. A claim that 97 to 98 percent of inquiries receive a first response within three minutes is likewise self-reported — and, to the author’s credit, immediately qualified with the admission that actual resolution may take considerably longer.
Two further data points deserve scrutiny. The first is a Visa figure stating that business and commercial use accounts for nearly 17 percent of stablecoin card transaction volume. That is a share, not a growth rate, yet the commentary deploys it as evidence that corporate adoption of crypto payments is rising. A 17 percent share is entirely compatible with a shrinking market. Proving growth requires a year-over-year change, not a proportion.
The second is temporal. The article’s citations point to an October 2026 social media post and a dataset spanning September 2025 to August 2026. If the piece was published before those dates, every statistic in it is either fabricated or forward-projected. The verification burden here is not academic; it determines whether any of the numbers can be cited at all.
The Missing Control Group
The most serious methodological gap is not the sourcing. It is the absence of a comparison.
The article’s central claim is that crypto payments carry hidden costs. To establish that those costs are distinctive — or even that they are higher than the alternative — requires a baseline. What is the failure rate on card transactions? What does a card chargeback actually cost a merchant in staff time, evidence assembly and lost revenue? What is the median resolution time for a card dispute?
None of this appears. The $20-per-failed-payment figure is itself drawn from banking research, not crypto, yet it is never set against a crypto equivalent. Without a control group, the proposition that crypto payments are more expensive on a total-cost basis is not merely unproven — it is unstated. The article demonstrates that crypto acceptance has operational costs. It does not demonstrate that those costs exceed the costs of the payment rails merchants already use.
The Technical Claim That Gives the Game Away
Buried in the middle of the piece is a detail that deserves more attention than the headline statistics: the assertion that the gateway can automatically handle misdirected deposits between compatible chains.
Read carefully, this sentence carries three implications. First, “compatible” is doing heavy lifting — only an asset supported on both chains, such as a dollar stablecoin on two networks, can be recovered this way. A deposit of a long-tail token the gateway does not list, or an asset sent to an address on an unsupported network, is effectively unrecoverable, because cross-chain recovery requires control of the destination private key.
Second, that is precisely the point: automated recovery is only possible if the service holds the keys to the receiving address, or has the ability to move funds out of that chain. That is custodial architecture, whatever the surrounding marketing language says about non-custody.
Third, the boundary of “compatible” is not published. Merchants cannot predict which errors will be rescued and which will be permanent, which is an information asymmetry with a direct financial cost.
The article also groups destination tag omissions alongside wrong-chain and wrong-asset errors as common failure modes. They are not comparable. A missing memo affects memo-based ledgers such as XRP, Stellar, Cosmos and TON. It is largely irrelevant on EVM chains and on Bitcoin’s UTXO model. The three error classes have different probabilities, different recoverability and different liability profiles; treating them as one category inflates the apparent failure rate.
What the Article Doesn’t Say
Three omissions stand out.
Compliance. There is no mention of licensing, know-your-customer obligations, sanctions screening or custody arrangements for funds in transit. For a business that holds merchant funds mid-flight, these are not ancillary details — they are the foundation of the operating model. The article’s suggested merchant checklist asks who tracks a problem, who issues the refund and who confirms resolution. It does not ask what licence the provider holds or how it handles a sanctioned address. That is a conspicuous boundary.
Self-custody. Open-source, self-hosted gateway software has existed since 2017, charges no processing fee and holds no keys. It is the natural counterpoint to a managed gateway: zero fee, but the merchant absorbs all operational burden. Its complete absence from a piece advising merchants on how to choose a gateway leaves the recommendation with exactly one available answer.
Accounting and tax. Crypto acceptance requires fair-value recording at the point of sale, cost-basis tracking across every transaction and jurisdiction-specific treatment that in many places classifies crypto payment as barter rather than settlement. For a growing merchant, this can dwarf support hours. The article confines “hidden cost” to customer service time.
The Structural Threat Nobody Is Pricing
The commentary’s own data source hints at a larger problem for standalone gateways. If business use already represents a meaningful share of stablecoin card volume, the dominant path for stablecoin commerce may not be merchant-to-gateway at all. It may be stablecoin to card network to merchant — with the crypto leg absorbed into existing settlement infrastructure.
In that scenario, independent gateways are demoted to on-ramp and off-ramp plumbing, and the pricing power migrates to the networks that already own the merchant relationship. Card networks and incumbent processors are positioned to capture the fee, while the crypto-native layer handles conversion.
That is a structural downgrade, and it is not discussed.
What Merchants Should Ask
The article’s three questions — who tracks the problem, who issues the refund, who confirms it is resolved — are a reasonable starting point. They need three more: what is your licence and where does your entity sit; what is your documented recovery rate and median resolution time, in writing; and how does your pricing compare with a self-hosted alternative we run ourselves.
The commentary’s contribution is the reframing, not the conclusion. Moving merchant attention from fee percentage to cost per exception is genuinely useful. But a correct framework filled with vendor-supplied, unverified figures and no comparison group cannot support a procurement decision.
The piece warns about hidden costs while concealing several of its own — compliance, fiat off-ramp friction, accounting overhead and counterparty risk. The irony is structural, and merchants should read it accordingly.
